Total Burn by Firebreak Type
Each scenario places 24–25 firebreak cells in a different configuration: a ridgeline north of Geyserville, the Highway 101 corridor, defensive perimeters around Geyserville or Windsor, or an east-west cross-valley cut. All tested across 50 weather draws with ±30% wind speed, ±20° direction, and ±15% humidity perturbation.
The Highway 101 corridor achieves the largest total burn reduction at 1.1% (mean 2,409 vs. baseline 2,435 cells). But total burn is not the right metric for community protection — arrival time and reach probability are.
Model: 50×50 Rothermel CA grid, 100m cells, 50 MC draws per scenario, wind/humidity perturbation. Runtime: 516s total.
What Matters: Arrival Time & Reach Probability
Total burn barely changes — fire finds a way around every break. The real question is whether a firebreak buys enough time for evacuation or prevents fire from reaching a community at all.
The Geyserville perimeter drops Geyserville's reach probability from 100% to 94% and pushes median arrival from 29 to 47 hours — an 18-hour gain. The Windsor perimeter drops Windsor's reach probability to 96% and pushes median arrival from 46 to 53 hours. No scenario protects Healdsburg — it reaches 100% in every configuration.
Total area is the wrong metric. Firebreaks cut burn area by ~1% — irrelevant at the landscape scale. What matters: do they delay fire long enough for a specific community to evacuate? Geyserville's perimeter shifts arrival from 29 to 47 hours — 18 extra hours against a community that needs ~2 hours to clear. That's the difference between evacuating under fire and leaving with time to spare.
Compare the alternatives. A firebreak buys ~1% area reduction and protects one community from one direction. An MC-informed early warning system (Q4) buys 1–7 hours of evacuation time across every community, regardless of wind direction. For limited budgets, the warning system is higher leverage than the earthwork. The firebreak can't adapt to a wind shift. The MC trigger already accounts for it.
You don't need CFD to rank firebreak locations. The physics just needs to be directionally correct across many weather scenarios. 50 MC draws stabilizes the ranking. The absolute numbers are rough. The relative comparison is what the decision actually needs.
50 Monte Carlo draws is enough to rank these options. A single deterministic run can't distinguish a genuinely effective firebreak from one that happened to work under today's wind. 50 draws stabilizes the ranking. The question is "where to clear fuel" — you don't need to resolve flame chemistry to answer that.